
Scanning
A scanner captures image data one line or "scan" at a time. A light source moves over a grid of (x, y) coordinates, that define sample areas. The light is reflected by mirrors through the optics of the scanner to a sensor, which converts the light to an electrical charge. The strength of the charge is in direct proportion to the intensity of the light reflected from the source image. The electrical signal is converted to a digital signal using an analog-to-digital converter. This digital information makes up the image file data (which is organized according to the file format that you choose).
The Scanning Process| 1. | PREVIEW | Previewing lets you see a quick, small-scale version of the image before committing to the final scan. At this stage, you also see a frame or marquee around the image. You can adjust this frame to crop in on the portion of your image that you wish to scan. |
| 2. | CROP to size | Crop during the scanning process (as opposed to cropping the image in Photoshop), because it reduces the scanning time. Cropping also determines the way your image will look by confining the area in which the scanner's auto exposure senses the darkest and brihtest areas of the image. |
| 3. | SELECT INPUT MODE |
|
| 4. | DECIDE RESOLUTION & IMAGE SIZE | There is a trade-off between higher resolution (which produces a better quality image) and file size (limited by the computer's memory). A 8 x 10", 300 DPI, RGB image is 20.6 MB (with no layers). |
Scanning
101: Setting the Right Resolution
In a dither about dots, halftones, and line screens? Luisa
Simone gives you the skinny on how to choose the best scan resolution for your
output device.
By Luisa Simone (from www.creativepro.com)
Tuesday, August 15, 2000
Your first goal as a conscientious print designer is to create the highest quality
scanned images for your clients. But too often designers interpret this responsibility
as a mandate to scan original artwork at very high resolutions, which produces
large files that eat up disk space, are difficult to transfer to a service bureau,
and take a long time to process at output time. If you really want to do the
best job possible all the way around, then you should strive for efficiency
in your scanned images as well as quality. Here we'll help you do just that,
by focusing on the optimum resolution you should use when scanning images for
printed output.
To determine the best scanning resolution, you need to consider two factors -- the resolution of the final output device and the type of artwork you are scanning. For the purposes of this discussion, all artwork falls into two basic categories -- black-and-white line art and continuous-tone images.
One Equals One
Regardless of the actual color used, an image qualifies as black-and-white artwork
if it consists of only one color without any tonal variation. Line-art drawings
and technical illustrations are obviously black-and-white images. But a blueprint,
a purple-hued company logo, or a red text headline could also be scanned as
black-and-white images.
Scanning black-and-white originals is actually a fairly straightforward process because there is a one-to-one correspondence between the resolution of a scanned line-art image (measured in pixels per inch, or ppi) and the resolution of the printer (measured in dots per inch, or dpi). For example, if you intend to print a black-and-white line-art image on a 600 dpi laser printer, then you should scan at 600 ppi.
In general, you can increase the quality of a black-and-white scan by increasing the resolution. Figure 1 illustrates how your choice of resolution affects the final image. At low-resolutions, such as 75 ppi, black-and-white scans can exhibit serious artifacts, including broken lines and aliasing -- which refers to the jagged staircase pattern on curved or diagonal edges. At 600 ppi, those same lines remain unbroken and reproduce with much smoother edges.
You can keep your life very simple by always scanning line art images at the exact resolution of your output device, regardless of whether it is a desktop printer or an imagesetter at a service bureau. In practice, though, there is a limit beyond which increasing the resolution of a black-and-white image has no benefit, because the human eye cannot perceive the difference. So while image setters may be capable of generating black-and-white images at up to 2,540 dpi, scanning the extra data is a waste of time (especially processing time). Table 1 lists the hardware resolution of various output devices along with the optimum scan resolution for each.
If you were to inspect a photograph very closely -- with the aid of a magnifier, for instance -- you would see that what appears to be a solid color actually consists of thousands of variations of that color. Indeed, a color photograph can contain millions of colors. In the same way, a "black-and-white" photograph actually contains hundreds of shades of gray, with black and white being merely the beginning and ending points of the gray spectrum. When a scanner captures a photograph it must convert the continuous tone information into a discrete number of hues and gray shades. The trick is to capture as much tonal information as possible.
If you are scanning a black-and-white
photograph, for example, you should make sure you are working in grayscale mode
and are capturing a full 8-bits of tonal information per dot, which will yield
up to 256 shades of gray. If you are scanning color photographs you should capture
24-bit bits of data per pixel to reproduce the full spectrum of 16.7 million
colors of which an RGB monitor is capable. (Scanning at higher than 24 bits
may make sense for you. If you're feeling daring, check out Bruce Fraser's recent
article on using high-bit color.)
One word of warning: Your particular scanning utility may use arbitrary naming
conventions to designate different scanning modes. For example, many scanning
utilities slip back into common parlance and refer to black-and-white photographs
instead of grayscale pictures. It's a phrase that's easily confused with black-and-white
line art -- a totally different scanning mode discussed earlier in this piece.
And there are some older scanning utilities that use the term "Color Photograph"
to identify an indexed color scanning mode, which uses a fixed palette containing
only 256 colors. To be absolutely sure that you have chosen the correct scanning
mode, double-check the documentation for your particular scanner.
Going Half Way
Provided that you choose the right scanning mode, and capture 8-bit grayscale
images or 24-bit color images, you'll find that your computer monitor can handle
the subtle variations in tone and color, because each red, green, or blue pixel
on a monitor can display one of 256 levels of color intensity. Printing devices
do not have this capability.
Instead, printers simulate the 256 different grayscale values present in an 8-bit image using only black ink on white paper. Likewise, the multitude of color values in a 24-bit scanned picture must be reproduced using only four primary inks -- cyan, magenta, yellow, and black. Printing devices create the illusion of gray shades or color variation through the use of halftones (see sidebar, "Halftoning and Halftone Cells").
Things can get pretty confusing when you are trying to find a relationship between the resolution of a scanned image and the resolution of a printer. The most basic scanning rule states that the resolution of a scanned image should always be based on the capabilities of the output device. But when you are measuring a printer's ability to simulate gray shades or color variation, the output resolution is measured in the number of halftone lines per inch (lpi), not hardware dots per inch (dpi). So, when you read Table 2 (which lists the recommended scanning resolutions for different types of printing devices) pay special attention to the halftone resolution (or linescreen frequency) associated with each device.
Of course, there is no law
compelling you to use these standard line-screen frequencies. If you decide
to use a non-standard line-screen setting -- meaning one that is not summarized
in Table 2 -- you can compute the optimum scanning resolution with one of the
following two formulas:
Oversample Your Data
You may be wondering why the formulas given above require you to scan your photographs
at resolutions higher than the actual line-screen frequency used to print the
picture. The reason is oversampling. You may be familiar with oversampling as
a feature of your stereo CD player. The concept is the same whether applied
to digital music or digital images. Basically, oversampling gathers excess information
that allows the computer to average the value of a sampled sound (in the case
of a song) or a sampled dot (in the case of a picture), to arrive at a more-precise
value. When you scan a photograph at 1.5 or 2 times the line screen setting,
you are oversampling the data to ensure that the tonal and color values of the
scanned image are accurately translated into a halftone pattern. Here's the
important point: Choosing a resolution that is greater than 2 times the
line screen setting doesn't improve the accuracy of the halftoning process.
It simply wastes computing resources.
Life as a print designer is one big circle. You start with hard-copy images -- in the form of drawings or photographs. You scan them into a computer where you modify them with image-editing software or incorporate them into a DTP layout. And then you print the pictures back out to paper. By understanding the relationship between scanning resolution and output resolution, you can ensure that the circle of quality remains unbroken.